Mesoxylon Lomaxii and M. poroxyloides. 1013 
Fig. 5 ; PL LXXXIX, Fig. 18) : the phloem-zone with its large and con- 
spicuous secretory sacs (Figs. 6 and 7), the broad cortex, mostly of 
secondary origin, and the rather narrow external, Dictyoxylon zone (Figs. 1, 
2, and 6 ), which, however, is seldom preserved. 
The Pith. 
The persistent zone of the pith shows little differentiation ; the outer 
portion has somewhat thicker walls than the inner layers, which consist of 
a delicate tissue ; the cells of both portions are short. 
The diaphragms constituting the central, discoid part of the pith are 
made up of thin-walled cells, similar to those of the inner portion of the 
persistent zone ; they have the form usual in Cordaiteae ; the diaphragms 
are thickest at the outer end, where they abut on the persistent zone, and 
thin out towards the centre of the pith, where they are usually broken and 
distorted (see the radial section, PL LXXXIX, Fig. 17). 
The Leaf-traces. 
As regards the course of the leaf-trace strands our data are incomplete, 
for no serial sections are available ; the characteristic points, however, are 
clear. Traced inwards through the secondary wood the twin bundles of 
a leaf-trace converge (PL LXXXVII, Fig. 4), and fuse as soon as they 
reach the margin of the pith (P"ig. 5). This rapid convergence and fusion 
of the twin strands may be taken, provisionally, as a specific character. 
As a result, the numerous circum-medullary primary strands are, in the case 
of M. Lomaxii , single bundles (PL LXXXIX, Fig. 18), with perhaps some 
slight indication of recent fusion. The tangential section shown in Fig. 3 
passes through a pair of bundles (at/./.), where they have not yet fused. If 
we now follow the trace outwards, we find that its course approaches the 
horizontal in passing through the wood, as is necessarily the case, where 
much secondary growth has taken place. In the phloem the two bundles 
reappear in approximately transverse section (Fig. 6). In passing through 
the cortex further subdivision goes on. Thus the four bundles of a single 
trace are seen in Fig. 7. Later stages of subdivision are not clearly shown, 
but it is evident that the bundles of the trace increased in number, for at 
least five are seen in a tangential section through the secondary cortex (slide 
2376). Probably the ultimate number was eight, as in the other species. 
The outgoing leaf-traces observed are not numerous. Though the 
loss of tissue, both in the wood and cortex, has no doubt often led to their 
being missed, it is probable that the leaves to which they ran were really 
more scattered than in the three species, M. Sutcliffii , M. poroxyloides , 
and M. midtirame. At the most not more than four traces have been 
observed outside the wood, nor more than three in the wood itself, in 
